iSCSI for beginner

Introduction

 

If you want to know more about iSCSI, we’ll have to talk a bit about SCSI. That seems logical! SCSI stands for Small Computer System Interface and is a bus. Basically, that’s a protocol to carry data. Historically, SCSI has always been related to a lot of very different kind of devices (from floppy readers to scanners) but storage got a particular place in its history. For instance, the father of SCSI, Al Shugart, was an engineer specialized in hard drives. This protocol allows fast and smart data transfers but is sometimes a bit complicated, especially when we talk about connections: a lot of specific hardware is required.

iSCSI allows SCSI commands to be transported through any IP network. That’s where the small “I” comes from. Basically, it’s like carrying a letter (the SCSI command) all around the world within a plane (the iSCI protocol). You can choose domestic flights (local networks) or international flights (Internet) as a destination, it does not matter: you just need to buy the proper ticket (iSCSI configuration).
iSCSI for Beginner

Initiators and Targets: the essential concept

An iSCSI imitator is the device which is going to send the commands all over the network. It can be software of a particular piece of hardware. In daily usage, the initiator would be your computer which is going to “initiate” the connection with the iSCSI drive by giving various orders through the network.

So, if you followed correctly, you’ve already guess that the iSCSI target is your NAS! The target will receive several orders from the computer and send back the requested data. Configuring initiators and targets is a very easy task once you know where to go. To help you, Thecus provided some very efficient how-to guide in the Thecus Classroom on our website www.thecus.com.

But why should I use iSCSI?

That’s a good point. In storage, the essential point is giving orders and getting response from your server. To do so, with acceptable performances and safety, engineers don’t have that many solutions. IF you don’t want to do direct, classical and old-fashioned connections, you can use Fiber Channel or iSCSI. Fiber Channel is a very efficient way to connect your storage units in your network. But it needs specific wires, hardware and skills. It can really be deterrent for home usage or SMB. iSCSI brings a lot of advantages on a table: less costly, efficient, easy to start and so on. The main reason is iSCSI does not need specific network: just a basic IP address-based network should do the trick. And that mean it’s also working on the Internet! With proper configuration, you can create a target/initiator couple anywhere in the world.

Wait a second… what’s FCoE?

We talked a bit about Fiber Channel, saying it requires specific cables and hardware. Aware of this issue, people start developing FCoE. Basically, it’s like iSCSI but using Fiber Channel protocol with a different carrier (iSCSI uses TCP/IP while FCoE uses Ethernet). iSCSI is more targeted to home and SMB usages while FCoE seems more practical for enterprises and data centers. However, the struggle between those two standards is hard and so far pretty balanced. Adopting FCoE or iSCSI greatly depends on your skills, budget, existing infrastructure and if you got a doubt, feel free to ask an IT professional about it!

Youtube: Discover iSCSI and how to use it with a Thecus® NAS.

You may want to read:

How to Create an iSCSI Thin-Provision Target on Thecus NAS

How to Connect to an iSCSI Target Using Windows

13 Comments

  • 4 Nov 2012 | Permalink |

    Amazing! This blog looks just like my old one! It’s on a completely different topic but it has pretty much the same page layout and design. Excellent choice of colors!

  • 5 Nov 2012 | Permalink |

    Content is also very rich and I was like.

  • 8 Nov 2012 | Permalink |

    I’m not sure where you’re getting your information,
    but great topic. I needs to spend some time learning more
    or understanding more. Thanks for excellent info I was looking for
    this information for my mission.

  • 8 Nov 2012 | Permalink |

    great post, very informative. I wonder why the other specialists of this sector don’t notice this. You should continue your writing. I’m sure,
    you have a great readers’ base already!

  • 8 Nov 2012 | Permalink |

    I was suggested this web site by my cousin. I’m not sure whether this post is written by him as no one else know such detailed about my problem. You’re incredible!
    Thanks!

  • 8 Dec 2012 | Permalink |

    Excellent article. That’s exactly what I was looking for.

  • 12 Dec 2012 | Permalink |

    I just want to tell you that I am just beginner to blogging and site-building and absolutely enjoyed this web-site. Very likely I’m want to bookmark your site . You amazingly come with exceptional articles. Thanks for sharing with us your web-site.

    • 19 Dec 2012 | Permalink |

      I so agree with your post on this. I have vendors cattnonsly making the pitch about how we can save so much money by combining the data access networks together. My experience over the years makes me want to run screaming in terror. My FC environment is the most stable, solid environment; it never has an issue. It’s not that way because of any magical FC properties, but because it’s such a simple environment. My next stable environment is the private networks we use for iscsi and for Oracle RAC interconnects. Again they never get touched, never thought about, they just work; there are 2x switches there and they aren’t uplinked to the rest of the world (outside of mgmt).Now the regular IP ethernet network, that has constant oddness occurring, it’s extremely dynamic and variable. In one location with have probably 20-30 VLAN’s, multiple load balancers, firewalls, trunks, channels, etc. Not because ethernet is in itself forcing complexity upon us, but because of our internal needs it is; and I suspect that most ether networks of any size tend to be complex. This tends to be where the rubber hits the road on badness.I have some very recent examples of this: a network port is configured as a truck port (containing lots of VLAN’s), what would happen if someone puts a server in that port but doesn’t have an OS on it? System powers up, tries to PXE boot, timesout and reboots itself repeating the cycle over and over. Well what happened was a constant spanning tree event every 80 sec or so, preventing anybody from talking on *all* the VLAN’s (trunk port was participating in spanning tree). Another was when some new Cisco 10Gig DFC cards would start dropping MAC addresses from their cam tables. So randomly the switch would flip all the ports to flooding in the network. You could actually run tcpdump and catch unicast traffic turning our very expensive switches into network hubs causing massive packet loss because of network port overruns. We had Cisco case open for over a month trying to figure it out, eventually finding a VLAN with a lower bridge priority (don’t know why it only came into affect when we added the new DFC line cards in).These two examples generally can be handled by applications, but filesystems don’t like this. Delay a filesystem write for a while and see if it freaks out. Take hundreds of servers (maybe throw in virtualization and jump it up to thousands?) and think about what kind of a fun day you’d have trying to get all of them to back and happy after their filesystems have shutdown or flipped to readonly mode because they couldn’t write to their filesystem for 30sec while rapid spanning tree kicked in.I will add that having a A B network with full physical separation and load balancing software like most FC implementations have would help reduce this issue, but network people aren’t doing that configuration (as a standard). They are bringing in redundant network connections from different switches, but they are physically connected together and the host runs lacp or active/passive connections and the host shares an ip for both connections. For full physical ether separation to ride through the problems I had above you’d need to have separate host ip’s and networks and some scsi layer load balancing software. None of the FCOE vendors are pitching this and the network people aren’t liking the idea of physically separate identical network infrastructure. When this becomes the standard for infrastructure then I’ll be willing to share block storage with general traffic on an ether network link.

  • 16 Dec 2012 | Permalink |

    This post is worth everyone’s attention. When can I find out more?

  • 17 Dec 2012 | Permalink |

    I just like the treasured details you present within your article content.I will bookmark your blog page and check once again right here commonly.I’m very certainly I will understand a whole lot of new stuff suitable here! Beneficial luck for that subsequent!

  • 22 Dec 2012 | Permalink |

    Well written post. Looking forward for more posts like this.

  • 10 Jan 2013 | Permalink |

    Hello, for all time i used to check web site
    posts here in the early hours in the morning, for the reason that i enjoy
    to gain knowledge of more and more.

  • 10 Jan 2013 | Permalink |

    Professional Tips For Wise law firm search engine optimization Secrets

One Trackback

  • 16 Nov 2017 | Permalink |

    Google

    Very handful of websites that happen to become comprehensive beneath, from our point of view are undoubtedly very well worth checking out.

Leave a comment

You must be logged in to post a comment.